The recent technological leap in AI capacities can be expected to engender a transformative paradigm shift in social robotics. Multimodal AI’s will soon allow social robots to respond to human verbal commands with autonomously generated simulations of human actions. In fact, social robots can be expected to act in a context-adequate fashion and to generate novel intentions. 

Multimodal AI thus represent a major step towards the longstanding vision of social roboticists to integrate robots “everywhere” into our lives, “at work and at home,” and to “personalize” robots. Multimodal AI will be a  game-changer for the equally longstanding socio-cultural concerns arising with this technology: how will social robots, with their radically improved practical ad simulatory skills,  affect societies and individuals: our economies, social practices, human social and intimate relations? As the margin of ‘humans-only’ competences is dwindling, how will this affect our cultural and individual self-comprehension? 

We live “the robotic moment” in human cultural history where we need to determine “who we are and who we are willing to become” (Sherry Turkle)—and now we have entered a decisive phase. Many AI researchers clearly perceive the risks of generative AI. These risks amplify when AI’s embedded in social robots can act out their “hallucinations”.  But risk management for social robots goes far beyond safety considerations—when artificial agents act in the physical and symbolic space of human social interactions, risk management requires expertise in analyzing the complex domain of social practices and human experiences.